This paper investigates whether the characteristics of locally elected officials influenced excess mortality during the COVID-19 pandemic. Using data on Italy, one of the first countries to be severely affected, we examine whether mayoral education influenced municipal-level mortality outcomes. We estimate weekly excess mortality using official death statistics and a Bayesian hierarchical spatio-temporal model. To address endogeneity in political selection, we implement a close-election Regression Discontinuity Design. We find that college-educated mayors significantly reduced mortality during the first wave of the pandemic, by lowering both the likelihood of excess deaths and the excess mortality rate. These effects are not observed in the second wave, likely due to policy convergence and a stronger role played by national and regional institutions. Our design interprets education as a proxy for broader leadership traits, such as decision-making capacity under uncertainty. The findings underscore that political selection can have real demographic consequences, shaping population outcomes during crises.